NetBird is an open-source networking platform that builds encrypted WireGuard mesh networks between devices, servers and cloud environments. Peers connect directly with built-in NAT traversal, so teams avoid routing traffic through a central VPN gateway. Access is governed by zero-trust controls: SSO and MFA through existing identity providers, device posture checks and granular access policies, all managed from a web UI or REST API. Teams can self-host it or use NetBird’s managed cloud, and it’s often seen as an open-source alternative to Tailscale and traditional VPNs. It fits Vultr well because it pairs naturally with Vultr VPC. Teams can use NetBird to reach Vultr instances securely, including ones without public IPs, and to connect Vultr resources with on-prem and multi-cloud environments. For the Alliance, its open-source core and self-hosting option reflect the same commitment to open, portable infrastructure, and it opens the door to joint solutions such as reference architectures and one-click deployments on Vultr.

Upbound is the company behind Crossplane, the CNCF open-source framework that extends Kubernetes into a control plane for managing cloud infrastructure and building internal developer platforms through declarative APIs. In June 2026 it released Modelplane, an Apache 2.0 control plane for AI inference built on Crossplane. It’s an early developer release that provisions GPU clusters, schedules model replicas across multi-cluster fleets, caches model weights and routes requests through a unified gateway, working with serving engines such as vLLM and SGLang. Both fit Vultr well. Crossplane maps onto our open APIs and Terraform-native automation, so teams can manage compute, Kubernetes and storage from one control plane. Modelplane lines up with Vultr Cloud GPU and our inference story, running across cloud, neocloud and on-premise environments. Upbound plans to donate Modelplane to a neutral foundation, which fits the Alliance’s open governance message and opens the door to joint reference architectures for platform engineering and AI inference.

Valkey is the Linux Foundation-governed, BSD-licensed fork of Redis 7.2, created in 2024 after Redis changed its license. It’s an in-memory data store for caching, session storage, real-time analytics, messaging and streaming. Valkey 8.0 added multi-threaded I/O for higher throughput, and it stays compatible with existing Redis clients and is backed by major cloud providers. Vultr already offers it as Vultr Managed Databases for Valkey, renamed from Managed Databases for Caching in January 2025 and backward compatible with existing setups. Teams can deploy CPU- or memory-optimized clusters in any Vultr location through the console, API, CLI or Terraform. They can attach up to three read-only replicas, including in other locations, and restrict access with VPC attachment, trusted sources and SSL. For the Alliance, Valkey is open governance in practice. An open-source engine delivered as a managed Vultr service gives us a ready joint story and a natural tie-in to the Linux Foundation relationship.

HAMi (Heterogeneous AI Computing Virtualization Middleware) is an open-source project that lets teams share GPUs on Kubernetes, so one GPU can serve several workloads instead of sitting mostly idle. It joined the CNCF in 2024 and was accepted as a CNCF Incubating project in July 2026. Workloads request a slice of GPU memory and compute through standard Kubernetes resource limits, HAMi enforces those limits at runtime, and its scheduler places pods across GPUs using policies such as topology awareness. It supports NVIDIA plus a growing list of other accelerators and works with schedulers such as Volcano and Kueue. It fits Vultr Cloud GPU and Kubernetes: teams can run HAMi on Vultr to put smaller inference, fine-tuning and dev workloads on shared GPUs, raising utilization and lowering cost per workload. For the Alliance, it’s a CNCF-governed, vendor-neutral way to get more from GPU capacity, and it opens the door to a joint reference architecture for GPU sharing on Vultr Kubernetes.

Qdrant is an open-source vector database and search engine, written in Rust and licensed under Apache 2.0, that powers semantic search, recommendations and retrieval-augmented generation (RAG). It uses HNSW indexing, combines dense and sparse vectors for hybrid search, and filters on JSON payloads, including keyword, numeric range and geo conditions. Built-in quantization can cut memory use by up to 97%, and collections scale horizontally through sharding and replication. Teams can self-host it, use Qdrant Cloud, or run Hybrid Cloud on their own Kubernetes cluster. It fits Vultr because AI teams need a vector store and the compute behind it. They can run Qdrant on Vultr Cloud Compute or Kubernetes and use Vultr Cloud GPU to generate embeddings and serve models. For the Alliance, Qdrant’s open-source core and Kubernetes-native deployment support portable AI stacks, and they open the door to joint reference architectures for RAG and AI search on Vultr.

Spinifex, from Mulga DC, is an open-source cloud platform (AGPL-3.0) that lets teams run the cloud services their applications already call on infrastructure they choose. It covers compute, block and object storage, VPC networking, identity, load balancing, managed Kubernetes, containers and databases, and works with existing CLI, SDK and Terraform workflows through an endpoint change. It fits Vultr directly. Spinifex deploys with one command onto one or three Vultr Cloud Compute VMs or Bare Metal servers, using the Vultr Terraform provider and a private Vultr VPC. It provisions Vultr reserved IPs for public addresses automatically, and in a three-node cluster it restarts a failed node’s instances elsewhere, with their public IPs following. Teams keep their tooling while using Vultr’s global footprint. For the Alliance, it’s a practical, open-source migration path that opens the door to joint reference architectures, including sovereign and edge deployments on Vultr.